General deep learning framework for emissivity engineering

• DOI: 10.1038/s41377-023-01341-w
• Publication Date: 2023-12-05T09:02:15Z
• Authors: Shilv Yu; Peng Zhou; Wang Xi; Zihe Chen; Yuheng Deng; Xiaobing Luo; Wangnan Li; Junichiro Shiomi; Run Hu
• Keywords: Applied optics. Photonics; QC350-467; Optics. Light; Article; TA1501-1820

Wavelength-selective thermal emitters (WS-TEs) have been frequently designed to achieve desired target emissivity spectra, as a typical engineering, for broad applications such camouflage, radiative cooling, and gas sensing, etc. However, previous designs require prior knowledge of materials or structures different the WS-TEs usually vary from in terms structures, thus lacking general design framework engineering across applications. Moreover, fail tackle simultaneous both they either fix select suitable materials. Herein, we employ deep Q-learning network algorithm, reinforcement learning method based on framework, multilayer WS-TEs. To demonstrate validity, three are various applications, including cooling which then fabricated measured. The merits algorithm include that it can (1) offer beyond one-dimensional structures; (2) autonomously self-built material library (3) optimize structural parameters spectra. present is demonstrated be feasible efficient designing highly scalable materials, dimensions, functions, offering paving way nonlinear optimization problems metamaterials.